User praises Qwen3.6 27B with llama.cpp

A Reddit user shares a positive experience running Qwen3.6 27B with llama.cpp on ROCm, using flash-attn and a dual-GPU setup. The post includes a config example and highlights strong performance.